Multi response prediction of cutting force and delamination in carbon fiber reinforced polymer using backpropagation neural network-genetic algorithm

Abstract

Carbon Fiber Reinforced Polymer (CFRP) has been widely used in various industries, including automotive, trains, and especially aerospace as a substitute for metal materials because of its high specific strength. Milling is one of the most commonly used machining processes in composites. Thus, it is necessary to determine the exact machining process parameters so that the specifications of the components are met. Moreover, suitable optimization method is needed to obtain machining parameters that produce small delamination and low cutting force. Full factorial design (2x3x3) with spindle speed, cutting speed and depth of cut as an input and the responses of cutting force and delamination was carried out in this experiment. A genetic algorithm was used as an optimization method, while backpropagation neural networks (BPNN) were used to apply complex non-linear equations. The BPNN model optimized using the GA method has been successfully developed in which the end milling process on CFRP material gains a mean square error (MSE) of 0.0246. This value indicates that the BPNN-GA model can be used as a predictor of the end milling CFRP process and eventually used to optimize the machining process.